@uql/sqlite

{*} uql = Universal Query Language

uql is a plug & play ORM, with a declarative JSON syntax to query/update multiple data-sources. Essentially, you just declare what you want from the datasource, and then uql will run efficient (and safe) SQL (or Mongo) queries.

Given uql is just a small library with serializable JSON syntax, the queries can be written in the client (web/mobile) and send to the backend; or just use uql directly in the backend; or even use it in a mobile app with an embedded database.

:star2: Features

• readable and yet powerful API
• code is performant and secure
• supports on-demand populate (at multiple levels), projection of fields/columns (at multiple levels), complex filtering (at multiple levels), sorting, pagination, and more.
• declarative and programmatic transactions
• entity repositories
• different relations between the entities
• supports inheritance patterns between the entities
• connection pooling
• supports Postgres, MySQL, MariaDB, SQLite, MongoDB (more coming)
• plugins for frameworks: express (more coming)

Table of Contents

1. Installation
2. Configuration
3. Entities
4. Declarative Transactions
5. Programmatic Transactions
6. Generate REST APIs with Express
7. Consume REST APIs from the Frontend
8. FAQs

:battery: Installation

1. Install the core package:

   npm install @uql/core --save
   

   or

   yarn add @uql/core
   

2. Install one of database packages according to your database:

   • for MySQL (or MariaDB)

     npm install @uql/mysql --save
     

     or with yarn

     yarn add @uql/mysql
     

   • for PostgreSQL

     npm install @uql/postgres --save
     

     or with yarn

     yarn add @uql/postgres
     

   • for SQLite

     npm install @uql/sqlite --save
     

     or with yarn

     yarn add @uql/sqlite
     

   • for MongoDB

     npm install @uql/mongo --save
     

     or with yarn

     yarn add @uql/mongo
     

3. Your tsconfig.json needs the following flags:

"target": "es6", // or a more recent ecmascript version
"experimentalDecorators": true,
"emitDecoratorMetadata": true

:gear: Configuration

uql initialization should be done once (e.g. in one of the bootstrap files of your app).

import { setOptions } from '@uql/core';
import { PgQuerierPool } from '@uql/postgres';

setOptions({
  querierPool: new PgQuerierPool({
    host: 'localhost',
    user: 'theUser',
    password: 'thePassword',
    database: 'theDatabase',
  }),
  logger: console.log,
  debug: true,
});

:egg: Entities

Take any dump class (aka DTO) and annotate it with the decorators from '@uql/core/entity/decorator'.

Note: inheritance between entities is optional.

import { v4 as uuidv4 } from 'uuid';
import { Id, Property, OneToMany, OneToOne, ManyToOne, Entity } from '@uql/core/entity/decorator';

/**
 * an abstract class can (optionally) be used as a "template" for the entities
 * (so the common attributes' declaration is reused)
 */
export abstract class BaseEntity {
  @Id({ onInsert: () => uuidv4() })
  id?: string;

  /**
   * 'onInsert' callback can be used to specify a custom mechanism for
   * obtaining the value of a property when inserting:
   */
  @Property({ onInsert: () => Date.now() })
  createdAt?: number;

  /**
   * foreign-keys are really simple to specify
   */
  @Property({ reference: () => User })
  userId?: string;

  /**
   * 'onUpdate' callback can be used to specify a custom mechanism for
   * obtaining the value of a property when updating:
   */
  @Property({ onUpdate: () => Date.now() })
  updatedAt?: number;

  @Property()
  status?: number;
}

@Entity()
export class Company extends BaseEntity {
  @Property()
  name?: string;

  @Property()
  description?: string;
}

/**
 * a custom name can be specified for the corresponding table/document
 */
@Entity({ name: 'user_profile' })
export class Profile extends BaseEntity {
  /**
   * a custom name can be optionally specified for every property (this also overrides parent's class ID declaration)
   */
  @Id({ name: 'pk' })
  id?: string;

  @Property({ name: 'image' })
  picture?: string;

  /**
   * the owner side of the relation is that which doesn't specify the attribute `mappedBy`
   */
  @OneToOne()
  user?: User;
}

@Entity()
export class User extends BaseEntity {
  @Property()
  name?: string;

  @Property()
  email?: string;

  @Property()
  password?: string;

  /**
   * the non-owner side of the relation is that which specify the attribute `mappedBy`
   */
  @OneToOne({ entity: () => Profile, mappedBy: (profile) => profile.user })
  profile?: Profile;
}

@Entity()
export class TaxCategory extends BaseEntity {
  /**
   * an entity can specify its own ID Property and still inherit the others
   * columns/relations from its parent entity.
   * 'onInsert' callback can be used to specify a custom mechanism for
   * auto-generating the primary-key's value when inserting
   */
  @Id({ onInsert: () => uuidv4() })
  pk?: string;

  @Property()
  name?: string;

  @Property()
  description?: string;
}

@Entity()
export class Tax extends BaseEntity {
  @Property()
  name?: string;

  @Property()
  percentage?: number;

  @Property({ reference: () => TaxCategory })
  categoryId?: string;

  @ManyToOne()
  category?: TaxCategory;

  @Property()
  description?: string;
}

:speaking_head: Declarative Transactions

uql supports both, declarative and programmatic transactions, with the former you can just describe the scope of your transactions, with the later you have more flexibility (hence more responsibility).

1. take any service class, annotate the wanted function with the Transactional() decorator.
2. inject the querier instance by decorating one of the function's arguments with @InjectQuerier().

import { Querier } from '@uql/core/type';
import { Transactional, InjectQuerier } from '@uql/core/querier/decorator';

class ConfirmationService {
  @Transactional()
  async confirmAction(payload: Confirmation, @InjectQuerier() querier?: Querier) {
    if (payload.type === 'register') {
      await querier.insertOne(User, {
        name: payload.name,
        email: payload.email,
        password: payload.password,
      });
    } else {
      await querier.updateOneById(User, { password: payload.password }, payload.userId);
    }
    await querier.updateOneById(Confirmation, { status: CONFIRM_STATUS_VERIFIED }, payload.id);
  }
}

export const confirmationService = new ConfirmationService();

/**
 * then you could just import the constant `confirmationService` in another file,
 * and when you call `confirmAction` function, all the operations there
 * will (automatically) run inside a single transaction
 */
await confirmationService.confirmAction(data);

:hammer_and_wrench: Programmatic Transactions

uql supports both, declarative and programmatic transactions, with the former you can just describe the scope of your transactions, with the later you have more flexibility (hence more responsibility).

1. obtain the querier object with await getQuerier()
2. open a try/catch/finally block
3. start the transaction with await querier.beginTransaction()
4. perform the different operations using the querier or repositories
5. commit the transaction with await querier.commitTransaction()
6. in the catch block, add await querier.rollbackTransaction()
7. release the querier back to the pool with await querier.release() in the finally block.

import { getQuerier } from '@uql/core';

async function confirmAction(payload: Confirmation) {
  const querier = await getQuerier();
  try {
    await querier.beginTransaction();
    if (payload.action === 'signup') {
      await querier.insertOne(User, {
        name: payload.name,
        email: payload.email,
        password: payload.password,
      });
    } else {
      await querier.updateOneById(User, { password: payload.password }, payload.userId);
    }
    await querier.updateOneById(Confirmation, { status: CONFIRM_STATUS_VERIFIED }, payload.id);
    await querier.commitTransaction();
  } catch (error) {
    await querier.rollbackTransaction();
    throw error;
  } finally {
    await querier.release();
  }
}

:zap: Generate REST APIs from Express

uql provides a express plugin to automatically generate REST APIs for your entities.

1. Install express plugin in your server project:

npm install @uql/express --save

or with yarn

yarn add @uql/express

1. Initialize the express middleware in your server code to generate REST APIs for your entities

import * as express from 'express';
import { querierMiddleware } from '@uql/express';

const app = express();

app
  // ...
  .use(
    '/api',
    // this will generate REST APIs for the entities.
    querierMiddleware({
      // all entities will be automatically exposed unless
      // 'include' or 'exclude' options are provided
      exclude: [Confirmation],
    })
  );

:globe_with_meridians: Consume REST APIs from Frontend

uql provides a client plugin to consume the REST APIs from the frontend.

1. Install client plugin in your frontend project:

npm install @uql/client --save

or with yarn

yarn add @uql/client

1. Use the client to call the uql CRUD API

import { getRepository } from '@uql/client';

// 'User' is an entity class
const userRepository = getRepository(User);

const users = await userRepository.findMany({
  $filter: { name: { $startsWith: 'lorem' } },
  $populate: { profile: { $project: ['picture'] } },
  $sort: { createdAt: -1 },
  $limit: 100,
});

:book: Frequently Asked Questions

Why uql if there already are GraphQL, TypeORM, Mikro-ORM, Sequelize?

GraphQL requires additional servers and also learning a new language; uql should allow same this, but without need to configure and maintaining additional components.

On the other hand, existing ORMs like TypeORM, Mikro-ORM, and Sequelize; are in one way or another, coupled to databases; uql uses a declarative JSON syntax (agnostic from the datasource) which can easily be serialized and send as messages (e.g. through the network) between different components of a system (e.g. micro-services), and then each one has the flexibility to decide how to process these messages.

At last but not at least, uql helps with the communication between the different tiers of your system, e.g. it allows the frontend to send dynamic requests to the backend (like GraphQL).